Longitudinal Effects of Acute Cannabis Exposure on Automobile Driving Behavior in a Naturalistic Simulated Environment

Abstract

Background: Driving is a complex day-to-day activity that employs a variety of cognitive and psychomotor functions in harmony, many of which are known to be affected acutely by CNB intoxication which could in turn pose a significant public health risk. The recent legalization of both recreational and/or medicinal marijuana in several states has thus created an urgent need to better understand the effects of CNB on such functions in the context of driving. The present study employs a longitudinal, double-blind, placebo- 2 active dose study to investigate the effects of CNB on a variety of driving-related behaviors in a controlled, naturalistic simulated environment. Methods: The current study employed N=37 subjects (N=25 male, frequent cannabis users, mean age 24.25+7.01), each exposed to a placebo, low and high dose of CNB on three separate days. On each day, following a single acute inhaled 0.5 g dose of either 0%, 3% or 5-7% of THC via a desktop vaporizer, subjects drove a virtual driving simulator (RTI SimVehicle platform) three times inside an MRI scanner and once out of scanner, randomized, and dispersed throughout an eight hour daily period. During each driving session three distinct real time behavioral tasks corresponding to lane-keeping following simulated wind gusts (operational), lead car following (tactical) and safe overtaking (strategic) were assessed and corresponding behavioral data were computed using custom Matlab scripts. Data were analyzed using a mixed model framework in SPSS v24 which included dose, session, instrument (desktop v MRI), dose*session, dose*instrument and session*instrument as primary factors, covarying for age and sex. Results: Intoxicated subjects made significantly fewer gas pedal corrections (p<0.02) during the car following task and similarly fewer corrections to the steering reversal rate (p<0.02) during the lane weaving task, suggesting reduced awareness under the influence of cannabis. In addition we found that several variables showed significant differences in terms of estimates captured throughout the day suggesting that overall risk taking lessened as the day progressed and CNB effects wore off. Also, data trends suggested that under the high dose subjects took longer to return to baseline from their ‘impaired’ driving patterns. Key metrics that showed such significant daily effects included mean headway (p<0.001) and time to collision (p=0.02) from the car following task, deviation of lane position (p=0.03) from the lane weaving task, median gap (p=0.02) and overtaking speed (p=0.02) from the overtaking task. Although many driving measurements differed depending on whether driving was done in MRI or at a desktop setting, these differences had no relationship to different drug dose levels. Conclusion: In summary, key driving functions affected under higher doses of CNB largely agreed current cross sectional literature. Generally, largest impairments in driving behavior seemed to occur within 1-4 hours after drug exposure, which might have important implications for real life driving situations. Our preliminary analyses yield numerous metrics that changed throughout the day, suggesting broad-based impairment on many metrics commonly used to quantify driving performance and risk.